<snip
 Here is how I arrived at the cycle-to-cycle disparity.
 Assume 200HP@ 6000 RPM (10 ms/rev)
 Assume .40lb/hp/hr Assume 2 rotors with 2 injectors each
 Assume MOTR-04556 injectors
 Assume gal/hr x10.5 = cc/min
 From the spec sheet: turn-on time 1.14ms, turn-off time .85 ms

 200*.4 = 80lbs/hr = 840 cc/min divided by 2 rotors and 2 injectors/
 rotor = 210cc/min
 From Flow vs Voltage@100 hz a flow of 210cc/min requires a 3.5
 second pulsewidth. (I realize a 8ms period gives higher flow/minute)
 A 3.5ms pulse nets a flow time of prox 3.2ms ( 3.5 ms + 1.14ms turn-
 on - .85ms turn-off)
 At 6000 a train of 3.2 ms pulses repeated at 8 ms intervals will be
 partitioned into 10 ms intervals by the passing of the apex seal
 over the intake port.
 Assuming the first passing coincides with the arrival at the port of
 the leading edge of a fuel flow pulse(mark this instant as time T),
 intake will continue for 10 ms, ending 2 ms into the next fuel flow
 pulse, resulting in 5.2 ms of fuel flow into rotor chamber 1.
 Beginning at T+10ms,1.2 ms of flow remaining from the divided pulse
 will join 3.2 ms of flow from the pulse beginning at T+16ms, (ending
 at T+19.2). The charge into rotor chamber 2 is 4.4 ms of flow. The
 next fuel pulse begins at T+24ms and ends at T+27.2ms, and is the
 only fuel charged into rotor chamber 3, for 3.2 ms of fuel flow.
 Someone smarter than me can figure out the distribution of the
 percentage of times the leanest and richest events would occur, and
 the percentage of combustions which would be about right.

 My previous numbers of 4.8 and 3.2 were done by initiating parsing
 at the application of electrical pulse to the injector and not at
 the beginning of flow, which did not reveal the worst case
 disparity. My apologies.

 Victor Roberts

Wow, Victor, that was hard to follow.  I'll try to simplify the
concept and show the meat of the "problem" using a free-running
oscillator.  All that specific stuff about injector on/off times and
fuel delivery rates just muddies the concept.

I'm not sure why Paul doesn't get what Victor is saying.  Maybe Victor
and I aren't stating the problem simply enough.  Is anyone else
following this discussion having difficulties understanding what
Victor is worried about?  I'll try once more with diagrams.

For the record, I agree that there is a theoretical problem of chamber
charge disparity.  In reality I think induction mixing in the runners
will minimize this issue.  I have no evidence to back this up - it's
just my uneducated opinion.  :)

Here goes ... If your email uses proportional fonts the layouts below
will be jacked up.   For clarity, try cut-n-pasting it into notepad to
view in a monospace font.

Engine speed 6000RPM = 100rev/sec = 10ms/rev
Fixed pulse frequency 125Hz = 8ms/pulse
Below are three examples.  Shown are the resulting 40ms pictures of
chamber fuel charges; after 40ms the pattern just repeats.

Chamber#1  Chamber#2  Chamber#3  Chamber#4
--------- ---------- ---------- -----------
000000000 1111111111 2222222222 33333333334  Time in
123456789 0123456789 0123456789 01234567890  milliseconds

EXAMPLE #1: 12.5% duty cycle
---------- ---------- ---------- ----------  ------------------
1111111111 2222222222 3333333333 4444444444  Chamber #
F*******F* ******F*** ****F***** **F*******  F=Inj.on *=Inj.off
---------- ---------- ---------- ----------  ------------------
2ms=25%    1ms=12.5%  1ms=12.5%  1ms=12.5%   Chamber charges

EXAMPLE #2: 50% duty cycle
---------- ---------- ---------- ----------  ------------------
1111111111 2222222222 3333333333 4444444444  Chamber #
FFFF****FF FF****FFFF ****FFFF** **FFFF****  F=Inj.on *=Inj.off
---------- ---------- ---------- ----------  ------------------
6ms=60%    6ms=60%    4ms=40%    4ms=40%     Chamber charges

EXAMPLE #3: 75% duty cycle
---------- ---------- ---------- ----------  ------------------
1111111111 2222222222 3333333333 4444444444  Chamber #
FFFFFF**FF FFFF**FFFF FF**FFFFFF **FFFFFF**  F=Inj.on *=Inj.off
---------- ---------- ---------- ----------  ------------------
8ms=80%    8ms=80%    8ms=80%    6ms=60%     Chamber charges


I hope that helps, otherwise I'm throwing in the towel.  Especially
since 1) I don't think there will be a real-world issue, and 2) I
prefer microcontrollers.

-Vince Orton


Well done Vince. Good explanation. There are a bunch of airplanes flying
out there with the same system that backs you up. I think Tracy will admit his system is not timed.

Well when the analog circuitry starts getting complicated with other than linear functions I start preferring microcontrollers too. I was under the false impression a mass air flow sensor was linear.

Paul Lamar